Ecological restoration is essential to reverse land degradation worldwide. Most studies have assessed the restoration of ecosystem functions individually, as opposed to a holistic view. Here we developed a network-based ecosystem multifunctionality (EMF) framework to identify key functions in evaluating EMF restoration. Through synthesizing 293 restoration studies (2900 observations) following cropland abandonment, we found that individual soil functions played different roles in determining the restoration of belowground EMF. Soil carbon, total nitrogen and phosphatase were key functions to predict the recovery of belowground EMF. On average, abandoned cropland recovered about 19% of EMF during 18 years. The restoration of EMF became larger with longer recovery time and higher humidity index, but lower with increasing soil depth and initial soil carbon. Overall, this study presents a network-based EMF framework, effectively helping to evaluate the success of ecosystem restoration and identify the key functions.

We searched literature on the topic of 'Grain for Green program' in China and belowground functions during 1900-2019, by means of Web of Science, Google Scholar and China Knowledge Resource Integrated Database. This program aims to reduce water and soil erosion, enhance soil fertility, and recover degraded ecosystems. Restoration techniques include active and natural ecosystem restoration. We restricted our analyses to changes in soil C, N and P cycling processes, when croplands were restored to semi-natural ecosystems (i.e., forest, shrubland and grassland). The papers collected were further screened on the basis of the following criteria. First, both agricultural and restored sites shared similar climate and soil conditions. Second, studies were performed with paired sites, using chronosequence or retrospective design. Third, soil depth and restoration time were clearly described. Finally, belowground functions were explicitly shown by their mean values and sample sizes. Based on these criteria, we identified 293 studies from 272 papers that were suitable for our analysis. Table-form data were directly gathered from original papers, and figure-form data were digitized and extracted by the Engauge Digitizer (Free Software Foundation, Inc., Boston, MA). Overall, our dataset (2900 observations) spanned a mean annual precipitation range of 116 to 1980 mm and annual temperature range of -0.7 to 20.5 °C. Restored ecosystems included forests, shrublands and grasslands, with the restoration time from 1 to 100 years. Belowground processes represented multiple functions of storing nutrients (soil organic carbon-SOC, total N-TN, total P-TP), transforming nutrients (microbial biomass C-MBC, microbial biomass N-MBN, microbial biomass P-MBP, invertase, catalase, urease, phosphatase) and supplying nutrients (available N-AN, available P-AP).

(1) What else would someone need to know to use this dataset?

No

(2) Are there any missing values? 

No